Recently spoke to a knowledgeable Tech at IBM support
and was told that ADSM uses a FIFO approach to satisfying
multiple, concurrent requests for file I/O. Furthermore,
it has additional optimizations to reduce the number
of mount points required to service queued requests by
ordering the requests in a SSTR (Shortest Seek Time To Read)
fashion...

If anybody has additonal information on this topic, please
give it up, so I can capitalize on this information for the
design of a large and important system.

System Profile:
Here's what we've got driving our Magstar 3494...
AIX 4.3.2 running on a RS-6000 model H50
dual 366MHZ CPU, 2(usable) x 9.1GB Disk (18GB Total usable),
2GB RAM
ADSM version 2.4 the 3494 has an OS/2 box for the library mgr
and it is a Cryrix 166 MHZ CPU with 64 MB RAM

Our current storage requirements: I am projecting 1.2 - 3.6
million individual files and a total storage of 4.3TB


Also need good link(s) on performance benchmarks or results
if anybody knows of them.

Thanks in advance!
Chris